摘要
针对单一测度模型的特征评价方法存在特征敏感度"欠学习",以及支持向量机(support vector machines,简称SVM)参数优化算法普遍存在收敛速度慢、易陷入局部极值等问题,提出一种量子遗传算法优化的SVM滚动轴承故障诊断方法。首先,采集振动信号中的时域和频域特征构成多域多类别原始故障特征集;其次,构建一个基于相关性、距离及信息等测度的混合特征评价模型,得到特征权重与特征值组合构成的加权故障特征集;最后,将加权故障特征集为输入,将量子熵引入到量子遗传算法当中,对SVM的结构参数进行全局优化,完成滚动轴承故障模式的识别。试验结果表明,该方法能够以更快的速度收敛至全局最优解,在保证聚类性能的基础上提高了滚动轴承的诊断精度。
There is an"under-learning"problem in the feature sensitivity of feature evaluation method for a single measure model.The support vector machine(SVM)parameter optimization algorithm has the disadvantages of slow convergence rate and easy to fall into the local extreme.Rolling bearing fault diagnosis of SVM based on improved quantum genetic algorithm method is proposed.Firstly,the characteristics of time domain,frequency domain constitute multi-domain original fault feature set.Secondly,a weighted model feature evaluation model based on correlation,distance and information is constructed.Finally,the weighted fault feature set is taken as input,and the quantum entropy is introduced into the improved quantum genetic algorithm(IQGA)to optimize the structural parameters of SVM.The intelligent identification of rolling bearing failure mode is completed.The experimental results show that compared with the classical quantum genetic algorithm(CQGA)and genetic algorithm(GA),the proposed method can quickly converge to the global optimal solution and ensure the clustering performance,and improve the diagnostic accuracy of rolling bearing.
出处
《振动.测试与诊断》
EI
CSCD
北大核心
2018年第4期843-851,共9页
Journal of Vibration,Measurement & Diagnosis
基金
国家自然科学基金资助项目(51575143)
黑龙江省自然科学基金资助项目(E2016046)
关键词
特征敏感度
混合特征评价
量子遗传算法
支持向量机
滚动轴承故障诊断
characteristic sensitivity
hybrid feature evaluation
quantum genetic algorithm
support vector machine
rolling bearing fault diagnosis